← All models
Claude Sonnet 5
claude-sonnet-5anthropicAvailableReleased 2026-05-14Frontier coding and agent performance at roughly half the Opus price, and the sensible default for everyday work. 1M context, adaptive thinking, strong tool use.
To keep the service available, Claude Sonnet 5 and GPT-5.6 Luna each have their own concurrency limit. Sonnet 5 allows up to 10 in-flight requests and may queue briefly during busy periods; Luna allows up to 15 and returns 429 immediately once they are all in use, so retry after a moment.
Claude can only be billed from the paid Wallet. Credit and Reward balances cannot pay for Claude requests.
TextToolsThinkingThinking: low, medium, high, xhigh, max
Pricing comparison
Each route's consumer price, and what multiple of the official list price it charges. Requests start at the cheapest route and only move to the next one when it fails, so you normally pay the top rate here. Lowest ready quote is marked. USD per million tokens.
| Route | Rate | Input | Output | Cache read | Cache write |
|---|---|---|---|---|---|
| Official | 1x | $2.00/M | $10.00/M | $0.300/M | $4.00/M |
| QuartzLowest | 0.2x | $0.400/M | $2.00/M | $0.060/M | $0.750/M |
| Meridian | 0.35x | $0.700/M | $3.50/M | $0.105/M | $1.31/M |
| Basalt | 0.4x | $0.800/M | $4.00/M | $0.120/M | $1.50/M |
/v1/messagesAvailableGuides and comparisons
- Claude Sonnet 5 vs Opus 5: which one to run
- API error codes: what each one means
- Cheapest models by task: what each kind of work actually costs
- Choosing among the 12 Claude model ids
- Choosing a coding model: a filter, not a ranking
- The Claude 1M-token context window: what it is good for
- Running Claude Code against token-share
- Claude extended thinking: when it earns its tokens
- Claude Haiku 4.5 for cheap tasks: check the multiplier first
- Claude Sonnet 5 vs GPT 5.6 Sol: price, context, and the switching cost
- Claude Sonnet 5 vs Sonnet 4.6: context, output ceiling, price
- Estimating a bill before you run the job
- Grok's 500K context window against the 1M-token models
- Largest context models: who reaches 1M, and what it costs to use it
- Model pricing explained: how the multiplier produces your bill
- Models for batch processing: when the bill is the requirement
- Models for code review: what a diff review actually costs
- Models for large codebases: when context size is the deciding factor
- OpenAI-compatible API setup: your first request
- Prompt caching: cutting cost on a repeated prefix
- Rate limits and quotas: what actually bounds your traffic
- Stable lanes explained: what stable-* buys and what it costs
- Streaming responses across the three route shapes
- Switching from Anthropic direct: migrating the SDK
- Thinking models compared: who has it, and what it costs across providers
- Tool calling compared: 27 models, three request shapes